Research On Malicious Node Detection And Abnormal Data Detection Scheme For Wireless Sensor Networks

Wireless sensor network(WSN)is a multi-hop wireless network system composed of a plurality of sensor nodes in a self-organizing manner.The main purpose of the WSN system is to perceive the environmental information in the coverage area of the acquisition system,and transmit to the administrator after processing.Because the amount of data generated by the network system is huge and the sensor nodes have limited bandwidth,power,and computing capabilities,they are particularly vulnerable to harsh environments and attacks by malicious vandals,resulting in malicious nodes and emergencies in the sensor network system.And the existence of abnormal data will have a tremendous impact on the system services of the sensor network.The two main tasks involved in wireless sensor network security are detection of abnormal nodes and detection of abnormal data,which are independent and complementary.In order to solve the above two problems,this thesis first proposes a dual-threshold sensor network malicious node detection scheme.Each sensor node maintains the trust value of the neighboring node to reflect its past behavior in the decision.The two thresholds are used to reduce the false positive rate and improve the accuracy of the event area detection,thereby detecting the malicious node more accurate without sacrificing the normal node.Simulation results show that the scheme with double threshold is better than the scheme with single threshold.After screening out the malicious nodes in the sensor network,an abnormal data detection method based on multi-layer distributed wireless sensor network is proposed.The scheme performs clustering,merging,and KNNbased anomaly cluster detection at the gateway node,and returns the abnormal cluster information to the underlying node for local detection.Through simulations on the datasets collected by the artificial dataset and the real environment,it is proved that the improved anomaly detection scheme can achieve high detection rate,low false alarm rate and lower the communication consumption compared to the centralized scheme and the reference scheme.However,the data distribution in the wireless sensor network maintains correlation in time and space.In order to make full use of its space-time correlation principle,a new abnormal data detection scheme is proposed.The program consists of two main procedures,self-test and weighted median detection.In the self-test,future data values are estimated using a time correlation based Kalman filter.Using the self-test results,weighted median detection is further implemented based on spatial correlation.When the measured data deviates from the median value of the weighted neighborhood data vector by more than a certain threshold,it is likely to be faulty.And it is discarded locally only if both processes are determined to be faulty.The simulation results show that the proposed algorithm outperforms two classical distributed detection algorithms in terms of detection accuracy and false positive rate.